Fortran toolkit

Borja Petit

correlation

function correlation(xvar1,xvar2,w,mask) result(corr)
  implicit none
  real(kind=8)            :: corr      ! output: correlation coefficient between xvar1 and xvar2
  real(kind=8)            :: xvar1(:)  ! input: first variable
  real(kind=8)            :: xvar2(:)  ! input: second variable
  real(kind=8) , optional :: w(:)      ! input: (optional) weights
  logical      , optional :: mask      ! input: (optional) retrictions on observations

This function returns the correlation coefficient between two (same size) vectors xvar1 and xvar2 given some (optional) weights w. If supplied, the vector w should have the same size as xvar1 and xvar2. If not supplied, the program assumes uniform weigthing.

The user can also supply a mask to compute the conditional correlation. The variable mask is a logical array of the same size of xvar1 and xvar2 indicating which observations should be considered.

Dependencies: varmean, varvar

(back to index)

Example

corr = correlation(xvar1,xvar2,mask = xvar1.gt.0.0d0 .and. xvar2.lt.5.0d0)

This computes the correlation between vectors xvar1 and xvar2 only for those pair of points that satify the condition xvar1}>0$ and xvar2}<5$.